Scheduling method of wind power generation for electricity market using state-of-charge transition and forecast error. Issue 1 (1st January 2019)

• Record Type: Journal Article
• Authors: Kikuchi, Aki; Ito, Masakazu; Hayashi, Yasuhiro
• Is Part Of: Journal of International Council on Electrical Engineering. Volume 9:Issue 1(2019)
• Page Start: 123
• Page End: 132
• Publication Date: 2019-01-01
• Subjects: Energy storage system -- forecast error -- generation scheduling -- machine learning -- state-of-charge transition -- wind power; Electrical engineering -- Periodicals; Electrical engineering; Periodicals; 621.3
• DOI: 10.1080/22348972.2020.1712018
• Languages: English
• ISSNs: 2233-5951

Abstract

ABSTRACT: This study focuses on the trade of power from wind power plants (WPPs) at an electricity market. To trade power, operators should produce generation schedules for bids and then supply the same amount of energy as the schedules. However, power from WPPs tends to fluctuate, which complicates scheduling. This paper proposes three scheduling methods with an energy storage system (ESS) to solve this problem. The first method considers state-of-charge (SOC) transition to maintain the appropriate SOC and minimize the imbalance between supplied and scheduled energy. The second and third methods consider SOC transition and forecast errors. The second method uses a linear regression model to estimate forecast errors. The third method adopts a bagged trees model, which is a machine learning method, to directly estimate the adjusted forecast data considering errors. Five patterns of the rated power of the ESS are assumed, and these three methods are simulated on each power. In comparison with the basic method, whose schedules are the same as the forecast, the third method can reduce 84% of the imbalance from the schedules when the rated power of the ESS is the minimum. The proposed methods help develop further correct and practical scheduling methods.